Let’s call them Large Language Models instead of gracing them with the exalted title “artificial intelligence.” Apparently, they have great potential. They can also be very annoying. For example, during a recent computer operating system upgrade, Macs incorporated LLM (large language model) technology into various word processing programs. Some people probably like it. It might save some wear and tear on your keyboard, I suppose. Here’s what happens: you’re innocently typing along and your LLM anticipates and autocompletes your words. I have to admit that, on the rare occasions that I text I find this helpful. I don’t text because I despise brief communiqués that are inevitably misunderstood. When I’m writing long form (my preference), I don’t like my computer guessing what I’m trying to say. Besides, I type faster than its suggestions most of the time.
We have gone after convenience over careful thought. How many times have I been made to feel bad because I’ve misunderstood a message thumbed in haste, or even an email sent as if it were a text? More than I care to count. LLMs have no feelings. They don’t understand what it is to be human, to be creative. Algorithms are only a small part of life. They have no place on a creative’s desktop. I even thought that I should choose a different word every single time just to see what this feisty algorithm might do. Even now I find that sometimes it has no idea where my thoughts are going. Creative people experience that themselves from time to time.
Certain sequences of words suggest the following word. I get that. The object of creative writing, however, is to subvert that in some way. If we knew just which way a novelist would go every time, why would we bother reading their books? LLMs thrive on predictability. They have no human experience of family tensions or heavy disappointments or unexpected elations. We, as a species tend to express ourselves in similar ways when such things happen, and certain words suggest themselves when a sequence of letters falls from our fingers. LLMs diminish us. They imply that our creative wordplay is but some kind of sequence of 0s and 1s that can be tamed and stored in a box. I suppose that for someone who has to write—say a work or school report—such thing might be a boon. It’s not, however, the intelligence that it claims to be.
Steve A. Wiggins 